AI “tastes” beer — then tells brewers how to make it better

AI-designed beers were preferred in a blind taste test.
Sign up for the Freethink Weekly newsletter!
A collection of our favorite stories straight to your inbox

AI that predicts how to improve the taste of beer could help brewers develop the next beloved brew — and avoid having their creations poured down the kitchen sink.

The beer challenge: All beers share the same basic ingredients — grain, hops, yeast, and water. But between the different varieties of the basic ingredients, different alcohol contents, and optional special ingredients, like fruits or spices, brewers have a near-infinite number of flavor combinations they can try when developing a new beer.

To help them know when they’ve landed on a winner, they might ask trained beer experts to judge their new creations, but tasting sessions can be expensive, and panelists’ opinions are subjective. Once the session is over, the brewer might know what the experts did or didn’t like, but they won’t necessarily know why or how they might be able to make a better beer.

Better beer: Researchers from KU Leuven in Belgium have now trained an AI to predict how much people will like a beer and how a recipe could be tweaked to increase drinkers’ enjoyment of it — information that could be hugely useful to brewers.

“Once we did blind tastings, the beers became better, and more generally appreciated.”

Kevin Verstrepen

To do this, they started by analyzing 250 different beers, measuring more than 200 chemical properties, including pH, alcohol content, and sugar concentration.

They then had 16 expert tasters review the beers, scoring them in 50 categories. Finally, they collected 180,000 reviews of the beers from the online platform RateBeer, which included scores for aroma, taste, and other attributes.

Armed with all of that data, the team trained an AI to find connections between a beer’s chemical properties and how much people would like it. 

The experiment: To test their system, the researchers then “spiked” several off-the-shelf commercial beers with the compounds that the AI predicted would make them tastier — and it worked.

“We had the models analyze these beers and then asked them ‘How can we make these beers better?’” lead study author Kevin Verstrepen told MIT Technology Review.

“Then we went in and actually made those changes to the beers by adding flavor compounds,” he continued. “And lo and behold — once we did blind tastings, the beers became better, and more generally appreciated.”

“It’s not taking away the art of making good beer in an artisanal way.”

Kevin Verstrepen

Looking ahead: The ability of the AI to improve the taste of non-alcoholic beers, which generally receive the lowest scores from reviewers, could be their biggest application, according to Verstrepen. One of the beers they improved in their experiment was non-alcoholic, so it seems it can be done.

The team is working with Anheuser-Busch InBev — the largest brewer in the world and co-sponsor of the study — to explore that idea. Verstrepen is also collaborating with Belgian startup Bar.On to create a device that produces custom beers on demand.

While Verstrepen believes his team’s AI could help make better beer, he doesn’t envision it ever replacing human brewers. 

“I really see it as a tool,” he told New Scientist. “It will be used especially to make better alcohol-free beer, but it’s not taking away the art of making good beer in an artisanal way.”

We’d love to hear from you! If you have a comment about this article or if you have a tip for a future Freethink story, please email us at [email protected].

Sign up for the Freethink Weekly newsletter!
A collection of our favorite stories straight to your inbox
Related
AI chatbots may ease the world’s loneliness (if they don’t make it worse)
AI chatbots may have certain advantages when roleplaying as our friends. They may also come with downsides that make our loneliness worse.
Will AI supercharge hacking — if it hasn’t already?
The future of hacking is coming at us fast, and it isn’t clear yet whether AI will help attackers and defenders more.
No, LLMs still can’t reason like humans. This simple test reveals why.
Most AI models are incredible at taking tests but easily bamboozled by basic reasoning. “Simple Bench” shows us why.
The future of fertility, from artificial wombs to AI-assisted IVF
A look back at the history of infertility treatments and ahead to the tech that could change everything we thought we knew about reproduction.
“Model collapse” threatens to kill progress on generative AIs
Generative AIs start churning out nonsense when trained on synthetic data — a problem that could put a ceiling on their ability to improve.
Up Next
A bipedal robot walking through a ditch in a forested area, stirring up dust as it moves.
Subscribe to Freethink for more great stories